Case Outstanding Development Method

Study guide: Exam 5

Case Outstanding Development Method projects future payments from the case outstanding balance rather than from paid or reported losses — developing the reserve the claims department has already set, instead of the losses that have emerged.

$U = \text{Paid to date} + \text{Case Outstanding} \times f_{\text{case}}$

$f_{\text{case}} = \frac{\text{All future payments}}{\text{Current case outstanding}}$

$f_{\text{case}}$ converts the current case balance into all remaining payments — including payments on claims not yet reported. It is estimated from mature years: for each historical cohort, divide everything paid after a given valuation by the case outstanding at that valuation.
Friedland's fuller version works period by period, using two ratios estimated from history: paid in the next period ÷ case outstanding at the start, and case outstanding at the end ÷ case outstanding at the start. Rolling these forward generates a payment stream and runs the case balance down to zero.
The method suits books where case reserving is consistent and disciplined — case outstanding is then a genuine forecast of the remaining cost — and where paid patterns are distorted by changing settlement rates.
Its weakness is the mirror image: it depends entirely on case adequacy being stable. A strengthening inflates the base to which $f_{\text{case}}$ is applied, and the historical factor was estimated under the old, weaker regime.
It is the natural companion to a paid projection rather than a substitute: paid data and case data are independent inputs, so agreement between the two is meaningful evidence.
Do not develop paid losses to ultimate and then add developed case outstanding — that double counts. Paid-to-date is taken at face value; only the case balance is developed.

Projecting from Case Outstanding AY 2023 at $24$ months: paid to date $\$800{,}000$ , case outstanding $\$500{,}000$ , so reported is $\$1{,}300{,}000$ . From mature accident years, payments after $24$ months have averaged $2.00$ times the case outstanding at $24$ months.

Estimate ultimate losses and IBNR, and compare with a reported chain ladder at $\text{CDF} = 1.350$ .
Answer
 $\begin{align*} \text{Future payments} &= \$500{,}000 \times 2.00 \\ &= \$1{,}000{,}000 \\[4pt] U &= \$800{,}000 + \$1{,}000{,}000 \\ &= \$1{,}800{,}000 \\[4pt] \text{IBNR} &= \$1{,}800{,}000 - \$1{,}300{,}000 \\ &= \$500{,}000 \end{align*}$ 
Reported chain ladder:

$\$1{,}300{,}000 \times 1.350 = \$1{,}755{,}000$

The two are within $2.6\%$ — reassuring, and worth more than either estimate alone, because they use different information. The chain ladder uses the emergence pattern of reported losses; the case method uses only paid-to-date and the current reserve balance.

Note the paid figure is not developed. Adding a developed paid projection ( $\$800{,}000 \times 1.50 = \$1{,}200{,}000$ ) to the developed case balance would give $\$2{,}200{,}000$ and count the same future payments twice.

When Case Adequacy Shifts The same book strengthens case reserves by $30\%$ during $2024$ . At $12/31/2024$ , AY 2023 at $36$ months shows paid $\$1{,}100{,}000$ and case outstanding $\$650{,}000$ . The historical $36$ -month case factor, estimated from years reserved under the old regime, is $1.60$ .

What does the method give, and what is wrong with it?

Answer Applied mechanically:

$U = \$1{,}100{,}000 + \$650{,}000 \times 1.60 = \$2{,}140{,}000$

The problem: the $\$650{,}000$ case balance is $30\%$ stronger than the balances from which $1.60$ was estimated. Under the old regime the same claims would have carried about

$\frac{\$650{,}000}{1.30} = \$500{,}000$

and the historical factor is calibrated to that level of reserving. Applying $1.60$ to the strengthened balance builds the strengthening into the projection twice.

Two defensible fixes:

Restate the factor. If case reserves are now $30\%$ closer to their eventual settlement values, the future-payments-to-case ratio should fall correspondingly: roughly $1.60/1.30 = 1.23$ , giving

$U = \$1{,}100{,}000 + \$650{,}000 \times 1.23 = \$1{,}900{,}000$

Restate the history with the Berquist-Sherman case adjustment and re-estimate the factor from the restated triangle — more work, and more defensible, because it does not assume the adequacy change translates one-for-one into the ratio.

The general warning: every method that leans on case reserves — this one, the reported chain ladder, average case outstanding diagnostics — inherits whatever the claims department's reserving philosophy is doing.